Acute Pancreatitis in Patients With Severe Hypertriglyceridemia in a Multi-Ethnic Minority Population

Ambika Amblee, MD; Divyanshu Mohananey, MD; Micheal Morkos, MD; Sanjib Basu, PhD; Ayokunle T. Abegunde, MD; Malini Ganesh, MD; Neil Bhalerao, MD; Amrutha Mary George, MD; Milli Jain, MD; Leon Fogelfeld MD


Endocr Pract. 2018;24(5):429-436. 

In This Article


Based on the inclusion criteria, a total of 1,157 patients with TGs ≥1,000 mg/dL were included in the study; 871 patients had TGs 1,000 to 1,999 mg/dL, and 286 patients had TGs ≥2,000 mg/dL. The clinical and laboratory characteristics are shown in Table 1. Mean age was 49.2 ± 11.5 years, and males accounted for 75.6% of the study population. The majority (77.3%) consisted of ethnic minority populations: African American (31.6%), Hispanic (38.4%), Asian (5.7%), and Pacific Islander (1.6%). Excessive alcohol intake and history of smoking were found in 30 and 32.4%, respectively. A previous diagnosis of DM and HIV infection were found in 70.9 and 8.9%, respectively. Only 2% (24 patients) had history of gallstones.

The prevalence of AP in patients with severe HTG in this study was 9.2% (107 patients). Among patients who had AP, abdominal pain as the presenting symptom was found in 97.1%; 84.5% had a lipase level three times above the upper limit of normal, and 94.5% had CT evidence of AP. AP complicated by phlegmon formation occurred only in 2 patients, and both were HIV positive. The rate of AP was significantly less in HIV patients compared to patients without HIV (1.8% vs. 10.0%; P = .004).

Patients with AP were significantly younger than those without AP (41.3 years vs. 50.0 years; P<.001). AP was present in 13.7% of patients younger than 50 years, compared to 4.8% for those above 50 years of age (P<.001). Although men clearly outnumbered women in both groups, there was no significant difference in prevalence of AP between male and female patients (9.0% vs. 9.9%).

A history of excessive alcohol intake was found in 57.9% (90.3% males) of patients with AP, compared with 27.2% (86.7% males) of patients without AP (P<.001). In the AP group, excessive alcohol intake was found in 78.9% of Caucasians, 66.7% of African Americans, 42.9% of Asians, 48.9% of Hispanics, and 0.0% of Pacific Islanders. Excessive alcohol intake was significantly associated with HTG-AP in univariate analyses for all racial subgroups except Asians and Pacific Islanders.

History of gallstones was present in 6.5% of patients with AP and 1.62% of patients without AP (P = .004). Women were more likely to have gallstone disease than men (62.5% vs. 37.5%; P<.001). Smoking was equally prevalent in patients with and without AP (37.4% vs. 31.9%; P = .249). The DM prevalence and mean HbA1c values were comparable in patients with and without AP. Nine percent of the group had a diagnosis of HIV infection at the time of presentation. HIV patients were older than those without (54.5 ± 2.1 years vs. 41.1 ± 9.8 years; P = .05). The two patients with HIV and AP were African American males with TG levels of 1,785 and 2,265 mg/dL. Both HIV-positive patients had pseudocyst formation seen on imaging.

Patients with HTG-AP had higher incidence of diabetic ketoacidosis at admission (7.5% vs. 2.5%; P = .004); 88% were men and 12% were women in both cohorts. No one in our cohort had history of trauma, ERCP, pancreas divisum, autoimmune pancreatitis, or pregnancy.

In multivariable logistic regression analysis (Table 2), younger age, excessive alcohol intake, gallstone disease, and TGs >2,000 mg/dL remained significant independent risk factors for the development of HTG-AP.

The median TG level for patients with AP was higher than in those without AP (2,394 mg/dL vs. 1,406 mg/dL; P<.001). When patients were divided into two groups based on Endocrine Society guidelines for the classification of HTG (group 1, TGs between 1,000 and 1,999 mg/dL; group 2, TGs ≥2,000 mg/dL), the prevalence of AP was significantly higher in group 2, with 5.1% in group 1 and 22% in group 2 (Figure 1A). When patients were divided into four equal progressive TG level range groups, there was a significant and incremental increase in prevalence of AP from 6.6% in the lowest group to 66.6% in the highest TG group (Figure 1B).

Figure 1.

Prevalence of acute pancreatitis (AP) with and without presence of lifestyle risk factors. (A) Prevalence of AP in the two triglyceride (TG) groups. (B) Prevalence of AP in the four TG groups. (C) Prevalence of AP in the two TG groups with presence of 0, 1, or 2 risk factors (RF). *n is only 2, and hence, have to be cautious with this result. (D) Prevalence of AP in the four TG groups with presence of 0, 1, or 2 risk factors (rounded to the nearest decimal).

TG level association with AP was further stratified by the presence or absence of excessive alcohol intake and/or gallstones. Presence of one or two risk factors in each TG range group increased further the association with AP (Figure 1C and Figure 1D). It showed that patients in lower TG range groups (both in two or four range group analyses) and without any other risk factors have low association with AP (prevalence of 2.2% and 3.0%, respectively), and addition of even one risk factor raises the association significantly.

ROC Curve Analysis and Predictive Model

The ROC curve, using predicted values from a model with the four independent risk factors of age, TGs, presence or absence of excessive alcohol intake, and gallstones, yielded an area under the curve of 0.834 (95% CI, 0.80 to 0.87; P<.001) (Figure 2). The cross-validated ROC curve, using 3-fold cross-validated predictions, yielded a comparable area under the curve of 0.813 (95% CI, 0.77 to 0.85; P<.001).

Figure 2.

Receiver operating characteristic (ROC) curve analysis for the predictive model for hypertriglyceridemia-induced acute pancreatitis using the four independent risk factors of age, presence or absence of excessive alcohol intake, gallstones, and triglyceride level. Area under the curve: 0.834 (95% confidence interval, 0.80–0.87; P<.001).

A predictive model using the four independent risk factors was developed to calculate the probability of developing AP. Based on this model, we calculated several different scenarios of risk for two hypothetical patients: a younger 30 years old and an older 60 years old (Figure 3). The risk in the younger patient ranged from 1.7% and progressed as high as 57.0% as the number of risk factors increased. In the older patient, the risk range for AP was much lower (0.2% to 14.1%) as the same risk factors were accumulating. To access an interactive version of this model, visit

Figure 3.

Calculated risk for developing acute pancreatitis in two hypothetical patients (30 years old and 60 years old) with asymptomatic severe hypertriglyceridemia (triglyceride [TG] level >1,000 mg/dL) and different constellations of risk factors derived from the predictive model. A = alcohol; G = gallstones.

Based on our ROC curve, we suggest three cut-offs to determine the probability of developing AP (%): low risk as <4.4%, intermediate risk as 4.4% to <12.0%, and high risk as ≥12.0%. The sensitivity and specificity for the ≥4.4% cut-off were 94.4% and 52.9%, respectively, while for the ≥12.0% cut-off, they were 71.0% and 81.7%, respectively.